Structured Representation for Dynamic Survey Logic

School Name

Governor's School for Science & Mathematics

Grade Level

12th Grade

Presentation Topic

Computer Science

Presentation Type

Mentored

Mentor

Mentor: Marco Hirsch, German Research Center for Artificial Intelligence

Abstract

In this project, a survey system was designed which uses boolean logic to determine question order. Instead of moving linearly and branching at predefined positions in a survey, questions are chosen from a pool based on their relevances. This method allows the question selection to be more closely tailored to each respondent and can allow survey length to be reduced. This project describes a tool which parses an XML structure file and displays questions through a separate module. The system is divided into three parts. The main component is a central engine which parses and processes the survey question relevances to output a list of questions to display. The engine relies on a sensors module which collects data to be used in relevance calculation. The final component of the system is a display module written for a target platform which processes the output from the engine and displays it for the respondent. The relevances of questions can be defined using a simple, language-independent math syntax parsed by the tool or using virtual sensors written in any supported language for more flexibility, which allows any type of data to be used to calculate question relevance. The tool defined in this project will improve data collection by allowing survey designers to bypass redundant or irrelevant questions.

Start Date

3-25-2017 11:59 PM

Presentation Format

Written Only

Group Project

No

COinS
 
Mar 25th, 11:59 PM

Structured Representation for Dynamic Survey Logic

In this project, a survey system was designed which uses boolean logic to determine question order. Instead of moving linearly and branching at predefined positions in a survey, questions are chosen from a pool based on their relevances. This method allows the question selection to be more closely tailored to each respondent and can allow survey length to be reduced. This project describes a tool which parses an XML structure file and displays questions through a separate module. The system is divided into three parts. The main component is a central engine which parses and processes the survey question relevances to output a list of questions to display. The engine relies on a sensors module which collects data to be used in relevance calculation. The final component of the system is a display module written for a target platform which processes the output from the engine and displays it for the respondent. The relevances of questions can be defined using a simple, language-independent math syntax parsed by the tool or using virtual sensors written in any supported language for more flexibility, which allows any type of data to be used to calculate question relevance. The tool defined in this project will improve data collection by allowing survey designers to bypass redundant or irrelevant questions.